RESEARCH ARTICLE

Post-transcriptional regulation mediated by specific neurofilamentintrons in vivoChen Wang and Ben G. Szaro*

ABSTRACTNeurons regulate genes post-transcriptionally to coordinate thesupply of cytoskeletal proteins, such as the medium neurofilament(NEFM), with demand for structural materials in response toextracellular cues encountered by developing axons. By using amethod for evaluating functionality of cis-regulatory gene elementsin vivo through plasmid injection into Xenopus embryos, wediscovered that splicing of a specific nefm intron was required forrobust transgene expression, regardless of promoter or cell type.Transgenes utilizing the nefm 3′-UTR but substituting other nefmintrons expressed little or no protein owing to defects in handlingof the messenger (m)RNA as opposed to transcription or splicing.Post-transcriptional events at multiple steps, but mainly duringnucleocytoplasmic export, contributed to these varied levels ofprotein expression. An intron of the β-globin gene was also able topromote expression in a manner identical to that of the nefm intron,implying a more general preference for certain introns in controllingnefm expression. These results expand our knowledge of intron-mediated gene expression to encompass neurofilaments, indicatingan additional layer of complexity in the control of a cytoskeletal geneneeded for developing and maintaining healthy axons.

KEY WORDS: Post-transcriptional regulation, Cis-regulatoryelements, Intron, Neurofilament, Xenopus laevis

INTRODUCTIONAxon outgrowth during nervous system development requires thecoordinated synthesis of multiple neuronal cytoskeletal proteins. Thiscoordinated synthesis is an ongoing dynamic process involvingchanges in gene expression that are tightly regulated at both thetranscriptional and post-transcriptional levels in response to spatialand temporal cues (reviewed in Szaro and Strong, 2011). Becausetheir subunit composition changes continually with each stage ofaxon development, the neurofilaments provide a prime example ofthis regulation. As the intermediate filaments of neurons and the mostabundant cytoskeletal polymer of vertebrate myelinated axons,neurofilaments contribute to the formation, consolidation andmaintenance of the overall neuronal cytoskeletal network. Properlyregulated neurofilament gene expression promotes axon outgrowthand controls axon caliber, whereas dysregulated expressioncontributes to the pathogenesis of neurodegenerative disease(reviewed in Szaro and Strong, 2010; Thyagarajan et al., 2007).In both rodents and Xenopus, expression of the three type IV

neurofilament triplet subunit proteins [i.e. light (NEFL), medium

(NEFM), and heavy (NEFH) neurofilament proteins] responds tocues encountered by growing axons (Schwartz et al., 1990;Undamatla and Szaro, 2001; Zhao and Szaro, 1995). For example,during developmental and regenerative axon outgrowth, increasedexpression that is stimulated by axons making appropriate contactsis principally due to dynamic changes in nucleocytoplasmic export,stability, and translation of neurofilament RNAs, as opposed tochanges in gene transcription (Ananthakrishnan et al., 2008;Ananthakrishnan and Szaro, 2009; Schwartz et al., 1994). Inmature neurons, post-transcriptional changes in neurofilamentmessenger (m)RNA metabolism also contribute significantly toneurodegeneration (Cañete-Soler et al., 1999; Ge et al., 2003; Nieet al., 2002). To date, studies of the molecular mechanismsresponsible for these post-transcriptional changes have focusedmainly on identifying trans-factors [e.g. RNA-binding proteins andmicro (mi)RNAs] that bind to neurofilament RNAs. At least threeof the proteins that bind to neurofilament mRNAs – hnRNP K(Thyagarajan and Szaro, 2004, 2008), TDP-43 (Strong et al., 2007)and FUS/TLS (Lagier-Tourenne et al., 2012) – have importantfunctions in development and disease. For example, regulation ofneurofilament and other cytoskeletal-related mRNAs by hnRNP Kis specifically required for axon outgrowth (Liu et al., 2008; Liu andSzaro, 2011), and all three RNA-binding proteins have beenimplicated more generally in motor neuron disease (Arai et al.,2006; Kwiatkowski et al., 2009; Lagier-Tourenne and Cleveland,2009; Moujalled et al., 2015). These proteins bind not only to the 3′untranslated regions (UTRs) of their target RNAs but also to theintrons of a variety of neuronal pre-mRNAs (Cao et al., 2012;Lagier-Tourenne et al., 2012), suggesting that these proteinsinfluence multiple phases of the lives of their targeted RNAs, andthat their associations with introns could themselves be coupledwith subsequent events in the trafficking and translation of themRNAs (Lagier-Tourenne et al., 2012). Introns are well known toinfluence gene expression at multiple levels (Le Hir et al., 2003), butwhat role intron splicing plays for neurofilament RNAs is unknown,especially within the context of the natural regulation ofneurofilament gene expression that takes place within an intactdeveloping vertebrate nervous system. Answering this question isimportant for our understanding of the regulation of neuronalcytoskeletal composition in development and also has broaderimplications for neurodegenerative diseases.

Our recently developed in vivo technique for testing the activityof cis-regulatory gene regions in F0 generation Xenopus laevis, byinjecting plasmids expressing fluorescent reporter genes into two-cell-stage embryos (Wang and Szaro, 2015), provides an idealexperimental system for addressing the role of introns inneurofilament gene expression in an intact developing vertebrate.Of the neurofilament triplet genes, we chose to analyze the nefmgene, because its early expression in neural development (Bennettet al., 1988; Carden et al., 1987; Szaro et al., 1989), two-intron genestructure (Roosa et al., 2000) and 3′-UTR sequences (ThyagarajanReceived 18 December 2015; Accepted 17 February 2016

Department of Biological Sciences and the Center for Neuroscience Research,University at Albany, State University of New York, 1400 Washington Avenue,Albany, NY 12222, USA.

*Author for correspondence (bgs86@albany.edu)

1500

© 2016. Published by The Company of Biologists Ltd | Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

and Szaro, 2004) are phylogenetically highly conserved fromXenopus to mammals. We discovered that the nature of the intronitself, as well as splicing, is crucial for subsequent post-transcriptional events that culminate in protein expression. Ourfindings demonstrate the existence of an additional layer ofcomplexity in the control of cytoskeletal composition, whichneurons use to grow and maintain a healthy axon.

RESULTSThe second intron of the nefm gene promotes reporterprotein expression in vivoTo study the activity of cis-regulatory gene elements controllingpost-transcriptional regulation of the nefm RNA within thebiological context of a developing nervous system, we injectedplasmids bearing different cis-elements into Xenopus laevisembryos (Wang and Szaro, 2015). In this method, plasmidscontaining an attB element and two insulator sequences flankingthe reporter gene are injected into two-cell-stage embryos to yieldpromoter-specific reporter expression in F0 generation tadpoles thatpersists through at least early tadpole stages, thereby obviating theneed to create transgenic lines. Two fluorescent reporter constructs,a ‘test’ and a ‘reference’ plasmid, are co-injected unilaterally intothe embryos (Fig. 1). The test construct expresses a red fluorescentprotein (DsRed2) and contains the cis-regulatory regions of interest,whereas the reference plasmid expresses a green fluorescent protein(glGFP) to serve both as a control for selecting successfully injectedembryos and as a normalization factor in subsequent biochemical

analyses. In the current study, injected embryos were raised to latetailbud stages (stages 43–44), by which time they are completelytransparent, and Xenopus nefm exhibits a mature expression pattern(Szaro et al., 1989; Szaro and Gainer, 1988; Undamatla and Szaro,2001).

We initially set out to identify post-transcriptional cis-regulatoryelements of the nefm gene by analyzing the in vivo activity of its3′-UTR, because this region of the gene is highly conservedphylogenetically, contains multiple predicted binding sites formiRNAs, and binds to several RNA-binding proteins that post-transcriptionally regulate nefm expression (Liu et al., 2008;Thyagarajan and Szaro, 2004, 2008). For the reference plasmid,glGFPwas flanked by a 1.5 kb nefm promoter and a rabbit β-globin(hbb2) 3′-UTR, which have been demonstrated previously to yieldrobust reporter protein expression in neurons by stages 43–44(Wang and Szaro, 2015). For the test plasmid, the hbb2 3′-UTR andglGFP were replaced with the Xenopus nefm 3′-UTR and DsRed2,respectively (Fig. S1). When co-injected, the reference plasmidgave strong glGFP expression, confirming that procedures had beenperformed successfully, but the test plasmid generated very littleprotein (Fig. 2A, intronless), indicating that the nefm 3′-UTRneeded additional elements for optimal in vivo activity. Suchelements might lie anywhere upstream of the 3′-UTR, but becauseintrons are well known to enhance expression of reporter genes(Buchman and Berg, 1988; Callis et al., 1987; Duncker et al., 1997;Hamer et al., 1979; Le Hir et al., 2003; Palmiter et al., 1991), weopted to test nefm introns first.

Fig. 1. Microinjection of specialized plasmid DNAs into Xenopus embryos to analyze cis-regulatory gene element activity in vivo by usingimmunohistochemical and biochemical methods. (Left) A glGFP-expressing reference (top) was co-injected unilaterally into two-cell-stage embryos,together with various DsRed2-expressing test plasmids (e.g. bottom) bearing cis-elements of interest. Reference plasmid expression served as an internal controlfor the injection procedure, andas a normalizing factor for analyzing test plasmidRNAandprotein expression. Both plasmids possessed special elements (attBandinsulator sequences) for facilitating persistent promoter-specific expression in F0 animals (Wang and Szaro, 2015). For the purpose of demonstrating the use of themethod, both plasmids shown here are driven by an nefm promoter, restricting reporter expression to neuronal tissues.Χ indicates the injection site on a two-cell-stage embryo. (Middle) By stages 43–44 (late tailbud stages), tissue-specific expression of reporter proteins in cells that had descended from theinjected blastomere (left side of all images) could be visualized in live embryos through a fluorescence stereomicroscope. (Middle, top) Representative example ofthe anterior of a stage 43–44 embryo, previously co-injected with a glGFP reference plasmid and a DsRed2 test plasmid containing identical cis-regulatoryelements, under the control of a neuronal promoter (taken with a Nikon DS-Ri1 camera). *a.f., autofluorescence in the developing gut. (Right) Embryos exhibitingappropriate expression of the reference plasmidwere selected for further analyses of expressionwith double-label immunohistochemistry (green, glGFP;magenta,DsRed2) of cryosections through hindbrain and spinal cord (right, top) and by biochemical analyses of whole embryos (right, bottom). Immunostained transversecryosectionswere imagedwith aZeissLSM510confocal laser-scanningmicroscope.Dots outline the central nervous system,and the vertical dotted linesseparatethe injected (left) and uninjected sides. Scale bars: 20 µm (hindbrain); 50 µm (spinal cord).

1501

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

To test whether an upstream nefm intron would promote proteinexpression, we created new test plasmids by placing the last intron(Intron II) of Xenopus nefm near to the 3′ end of the DsRed2-codingsequence (Fig. S1), upstream of the stop codon and nefm 3′-UTR toprotect against nonsense-mediated decay (reviewed in Sun andMaquat, 2000). As determined by immunohistochemistical analysisof hindbrain and spinal cord, and by quantifying fluorescencein whole-embryo extracts on partially denaturing SDS 12%polyacrylamide gels (fluorescent SDS PAGE), addition of thisintron boosted protein expression by approximately threefold overthat of the intronless construct (Fig. 2A–C). To test whether theseeffects were specific to Intron II, we replaced it with the only otherintron in nefm, Intron I (Fig. S1). Protein expression with thisconstruct, however, was virtually undetectable (Fig. 3A,D,E; >14-fold less than Intron II), despite the intron having been successfullyspliced from expressed transcript (as determined by cloning andsequencing of the RNAs extracted directly from injected embryos).Thus, although nefm has two introns, only the splicing of Intron IIyielded optimal reporter protein expression.

Splicing alone is necessary but not sufficient for expressionTo test whether splicing of Intron II alone is sufficient to produce theobserved enhanced expression or whether the presence of additionalelements within this intron were also needed, we constructed twomore test plasmids (Fig. S1). The first had a minimal spliceablenefm Intron II, comprising 105 and 120 nucleotides from the 5′ and3′ ends, respectively, of the original full-length Intron II construct.Cloning and sequencing of the expressed RNA from injectedembryos confirmed that this intron was indeed spliced in vivo andthat the resultant RNA sequences were identical to those producedfrom the full-length Intron II construct. The second plasmid used anunspliceable nefm Intron II, which lacked 11 bp and 21 bp from the5′ and 3′ ends, respectively, of the full-length intron, eliminating thenative splice sites and rendering the intron unspliceable. In all cases,the co-injected reference plasmid yielded glGFP expression asusual, whereas neither test variant yielded detectable DsRed2protein (Fig. 3B–D). Quantification by performing fluorescent SDSPAGE indicated that the levels of DsRed2 protein from these two

variants, whether spliced or unspliced, were more than 20-fold lessthan that generated with the full-length Intron II (Fig. 3E).

Variations in protein expression profiles associatedwith theintrons are caused by differences in post-transcriptionalregulationTo gain insights into which control points in gene expression, fromtranscription to translation, are affected by having various introns,we assayed levels of reporter RNAs from different intracellularpools. Using primers targeting the reporter-coding regions, we firstmeasured the steady-state levels of reporter RNA from total cellularRNA, normalizing expression of DsRed2 to that of glGFP andexpressing results as a percentage of the expression of the intronlessconstruct group (Fig. 4A). The construct with the full-length IntronII generated the most reporter RNA, approximately threefold morethan the intronless group, which corresponded well with thethreefold increase seen in protein expression (Fig. 2C). Thus, thedifference in protein expression between these two groups could beaccounted for by a difference in the total amount of steady-stateRNA. However, differences in total RNA expression between theother intron groups and the full-length Intron II group were too smallto account for the much larger differences seen in proteinexpression. For example, the full-length Intron II constructyielded from ∼14 to >20-fold more protein than did the Intron I,minimal spliceable Intron II and unspliceable Intron II constructs(Fig. 3D,E), whereas it produced only 3.1-, 1.4- and 9-fold highersteady-state levels of total RNA, respectively (Fig. 4A). Thesedisparities between RNA and protein expression profiles suggestedthat additional regulatory mechanisms other than those that regulatesteady-state levels of total RNA contribute to variations in proteinexpression seen among the introns.

To test whether variations in RNA expression from differentintrons were due to differences in transcription, we performed real-time quantitative PCR (RT-qPCR) analyses to quantify nascentunspliced transcript by using primers spanning the 3′ splice-site ofthe introns to yield amplicons of the same length that could onlyhave been amplified from unspliced transcript. Quantification byperforming RT-qPCR demonstrated that all four intron-containing

Fig. 2. Reporter genes driven by the nefm promoter and paired with the nefm 3′-UTR yielded markedly stronger protein expression when the testplasmid contained the second intron of nefm (+nefm Int II) than when it was intronless. Embryos were injected with the indicated DsRed2-expressing testplasmid simultaneously with the glGFP reference plasmid (as in Fig. 1) and raised to stage 43–44 for analysis. (A) Double-label immunohistochemistry forglGFP (green) andDsRed2 (magenta) of representative sections fromhindbrain and spinal cord (40–50 sections from three animals each, for this and all subsequentfigures). Although both groups exhibited strong glGFP expression from the reference plasmid, only those receiving the nefm Intron II test construct expressedDsRed2 strongly. Dotted lines outline the central nervous system. Vertical dotted lines separate the injected and uninjected sides. Scale bars: 20 µm (hindbrain);50 µm (spinal cord). (B) Color: native fluorescence of DsRed2 (top band; test) and glGFP (bottom band; reference) from whole-embryo extracts, separated andvisualized directly on full-length partially denaturingSDS12%polyacrylamidegels (fluorescent SDSPAGE) to confirm differences in fluorescent reporter expression.DsRed2 runs as a dimer at ∼54 kDa and glGFP as a monomer at ∼27 kDa. Ladder: marker proteins running at ∼65, 50 and 30 kDa. Monochrome: imagesrepresentative of those used for quantification showing DsRed2 and glGFP bands. (C) Relative intensities of the DsRed2 fluorescent bands (normalized to glGFPfluorescence in the same lane) from the two groups further confirmed that expression was significantly greater for embryos expressing reporter genes containingnefm Intron II than those expressing the intronless counterparts. *P<0.01, Student’s t-test; n=3 replicates of 30 pooled embryos for each condition; error bars, s.e.m.

1502

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

groups generated equivalent amounts of nascent unspliced RNA(Fig. 4B), despite their having yielded widely varying levels of bothprotein and total RNA. Thus, the differences in steady-state levels ofreporter RNA among the four intron-containing groups arose notfrom differences in transcription, but rather from subsequentdifferences in RNA processing. Because such processing takesinto account the combined effects of pre-mRNA splicing andstability on mRNA levels, these differences were more readily anddirectly compared by encapsulating them into a single new statistic,termed mRNA processing efficiency (Fig. 4C). This statistic wascalculated by subtracting the ratio of nascent to total RNA from100%. As expected from the higher total RNA levels, the mRNAprocessing efficiency was modestly but nonetheless significantlygreater (P<0.01; Student’s t-test) for the full-length nefm Intron IIgroup than for the other three intron groups (1.3, 1.1 and 3.0 timesgreater than the Intron I, minimal spliceable Intron II and unsplicedIntron II groups, respectively). Thus, variations in the levels ofreporter RNA among these intron groups arose primarily fromdifferences in post-transcriptional rather than transcriptionalcontrol.Although differences in post-transcriptional control among the

four intron-containing groups were statistically significant, they

were insufficient to account for the larger differences seen in proteinexpression. One possibility was that the transcripts producedremained trapped within the nucleus, making them unavailablefor translation. To test this possibility, we compared the ratio ofcytoplasmic to nuclear reporter mRNA levels (Fig. 4D). Consistentwith other analyses, the full-length Intron II group exhibited thehighest ratio, with ∼10.4 times more DsRed2 mRNA in thecytoplasm than in the nucleus. In contrast, the Intron I and minimalspliced Intron II groups had less reporter RNA in the cytoplasm thanin the nucleus (52% and 77%, respectively; Fig. 4D). Moreover,there was no detectable unspliced Intron II RNA in the cytoplasm(thus unplotted). Consequently, the efficiency of export from thenucleus of the full-length Intron II transcript was 20- and 14-foldgreater than that of Intron I and the minimal spliced Intron II,respectively. From these data, we conclude that marked differencesin nucleocytoplasmic export of transcripts among the four intron-bearing constructs are the main contributor to the differences inprotein expression profiles seen in our system. In summary,accumulated differences in post-transcriptional regulation, arisingfrom diverse ways of handling the transcripts, ultimately make thenefm Intron II an optimal intron for promoting protein expressionwhen paired with the nefm 3′-UTR.

Fig. 3. Splicing was necessary, but not by itself sufficient, for robust protein expression of the nefm reporter gene. (A–C) Immunostained transversecryosections of hindbrain and spinal cord of stage 43–44 embryos that had been co-injected with the same reference plasmid as described in Fig. 2 and atest plasmid bearing either (A) a full-length nefm Intron I (+nefm Int I), (B) aminimal spliceable nefm Intron II (+nefm Int II mini spliced) –which contained 98 bp and100 bp from the 5′ and 3′ ends, respectively, of the original Intron II that had been used in the full-length Intron II construct, so that the spliced products derivedfrom these constructs would be identical, and (C) an unspliceable nefm Intron II (+nefm Int II unspliced) – which contained Intron II but lacked 11 and 20nucleotides from the 5′ and 3′ ends of the intron, respectively.Whereas robust glGFPexpression of the reference plasmid on the injected side (left of each section)indicated that experimental procedures were successful for all groups, DsRed2 expression of the test plasmids was barely visible in the same sections. Dottedlines outline the central nervous system. Vertical dotted lines separate the injected and uninjected sides. Scale bars: 20 µm (hindbrain); 50 µm (spinal cord).(D) Fluorescent SDS PAGE and (E) quantification of relative fluorescence intensities, which were performed, analyzed, and annotated as described in Fig. 2,confirmed that the full-length Intron II yielded significantly more DsRed2 expression than did Intron I and the two Intron II variants. *P<0.01 (Student’s t-test); n=3replicates of 30 pooled embryos for each condition; error bars, s.e.m.

1503

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

nefm Intron II works independently of the promoter andtissue type to promote protein expressionWe next tested the intronless construct and the full-length Intron IIconstruct with different promoters to assess whether this intronrequired specific promoter elements and expression in neurons tofunction. For these tests, we replaced the nefm promoter with twonew promoters – a neuronal β-tubulin promoter (tubb2b), whichlike the nefm promoter drives expression in neurons, and a cardiacactin promoter (actc1), which drives expression in muscle cells.For each new promoter, we created three plasmids: two DsRed2-expressing test plasmids (intronless and full-length nefm Intron II)both with the nefm 3′-UTR, and a glGFP-expressing referenceplasmid with the original hbb2 3′-UTR (Fig. S1). With both new

promoters, the nefm Intron II successfully promoted proteinexpression compared with its intronless counterpart – the tubb2bpromoter in neurons (Fig. 5A) and the atct1 promoter in musclecells (Fig. 5D). Quantification by performing fluorescent SDSPAGE (Fig. 5B,C,E,F) further demonstrated that this increasedexpression was both statistically significant (P<0.01, Student’st-test) and comparable in magnitude to that seen with the nefmpromoter (two- and fourfold greater for tubb2b and atct1,respectively, vs threefold for nefm). Thus, nefm Intron IIpromoted in vivo protein expression independently of bothpromoter and tissue type, further supporting the conclusion thateffects on protein expression are mediated post-transcriptionally(see Discussion).

Fig. 4. Variations in protein expression among introns arose from differences in post-transcriptional regulation operating at multiple control points.(A–C) Total RNAwas isolated from stage 43–44 embryos that had been previously co-injected with the glGFP reference plasmid and each of the indicatedDsRed2test plasmids. (A) Significant variations in reporter RNA expression were seen across all constructs, except between the intronless and nefm Intron I (nefm Int I)constructs. Expression of DsRed2 RNA relative to that of glGFP was determined by performing RT-qPCR (ΔCT) for each replicate (n=3 from 30 pooled embryos)using primers targeting sequences within each coding sequence for each reporter. For presentation, ΔCTwas converted to a fold change (2

–ΔCT) and then expressedas a percentage of the value of the intronless group. (B) All intron-bearing constructs gave rise to equivalent amounts of nascent RNAs. Levels of nascent unsplicedpre-mRNAs bearing the test reporter DsRed2 were determined by performing RT-qPCR analysis, using the same cDNAs that are described in A as templates butwith primers spanning 200 bp of the 3′ splice junctions. Forward primers targeted sites near to the 3′ end of each intron, and the reverse primer, which was thesame for all constructs, targeted the downstream 3′-UTR. DsRed2 nascent transcript levels are presented as a relative fold change over that of glGFP expression[2–ΔCT(DsRed2-glGFP)]. (C) mRNAs derived from Intron I and the two Intron II variants were subjected to less efficient post-transcriptional processing compared with full-length Intron II. Overall processing efficiencies of themRNAswere evaluated by taking the ratio of nascent unspliced RNAs given in B to the total RNA obtained in A,converting this ratio to a percentage, and subtracting the result from 100%. (D) Nucleocytoplasmic export of RNAs expressed from Intron I and the mini spliceableIntron II constructs was significantly hindered compared with that of the full-length Intron II. RT-qPCR analysis was performed separately on total RNA isolated fromnuclear and cytoplasmic subcellular fractions, using the same primers as used in A. ΔCT (cytoplasmic–nuclear) for DsRed2 was determined for each sample,averaged (mean±s.e.m.; n=3 biological replicates of ten embryos each), and converted to a fold change (2–ΔCT; i.e. cytoplasmic/nuclear). (A–D) *P<0.02; N.S.,P>0.05 (one-way ANOVA with Tukey’s post-hoc tests). Error bars (s.e.m.) are asymmetric owing to the conversion from ΔCT to 2–ΔCT (fold change). [Note, mRNAprocessing and export of the glGFP reference RNA were analyzed in similar ways and showed no significant differences across the same groups (Fig. S2)].

1504

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

The second intron of a rabbit β-globin gene and nefm Intron IIpromote equivalent expressionOur results raised the additional question of whether the introns and3′-UTRs of other genes display properties similar to those of nefmIntron II with its 3′-UTR. Closer inspection of the hbb2 3′-UTR ofthe reference plasmid revealed that it in fact contained a ∼0.6-kb-long intron (hbb2 Intron B), representing the second (and last)intron of the rabbit β-globin gene. In another context, the hbb2Intron B has been shown to promote both protein expression andnucleocytoplasmic mRNA export when included in a reporterconstruct transfected into BHK and COS cells (Rafiq et al., 1997).

Thus, we were further motivated to test whether the hbb2 Intron Bwould function in a similar way in our system. To test whether thisintron has the same effect on protein expression of test plasmidscontaining the hbb2 3′-UTR, we deleted hbb2 Intron B (Fig. S1)and found that protein expression diminished significantly, byapproximately threefold from that of the intron-bearing construct(Fig. 6). This difference was equivalent to that seen when the nefmIntron II was deleted from the construct comprising the nefm3′-UTR. Next, to test whether hbb2 Intron B promotes proteinexpression when coupled with the nefm 3′-UTR, we replaced nefmIntron II with hbb2 Intron B, while retaining the nefm 3′-UTR

Fig. 5. nefm Intron II increased protein expression independently of promoter and tissue type. Embryos were injected with constructs bearing either nointron (intronless) or the nefm Intron II (+nefm Int II) driven by either a neuronal β-tubulin (tubb2b) promoter, which drives expression in the nervous system (A–C),or a cardiac actin (actc1) promoter, which drives expression in muscle cells (D–F). A reference glGFP-expressing plasmid bearing the same promoter (tubb2b oractc1) as its corresponding test plasmid was co-injected in each case. Procedures, analyses and annotations were otherwise the same as described for Fig. 2.Dotted lines outline the central nervous system. Vertical dotted lines separate the injected and uninjected sides. Scale bars: 20 µm (hindbrain); 50 µm (spinalcord). *P<0.01 (Student’s t-test); n=3 replicates of 30 pooled embryos for each condition; error bars, s.e.m.

Fig. 6. When driven by the nefm promoter and paired with the rabbit β-globin (hbb2) 3′-UTR, hbb2 Intron B also increased reporter protein expressioncompared with its intronless counterpart. Embryos were co-injected with the same reference plasmid as in Fig. 2 and a test plasmid bearing either no intron(hbb2 3′-UTR+intronless) or the hbb2 Intron B (hbb2 3′-UTR+hbb2 Int B), all with the nefm promoter. Procedures, analyses and annotations were otherwise thesame as for Fig. 2. Dotted lines outline the central nervous system. Vertical dotted lines separate the injected and uninjected sides. Scale bars: 20 µm (hindbrain);50 µm (spinal cord). *P<0.01 (Student’s t-test); n=3 replicates of 30 pooled embryos for each condition; error bars, s.e.m.

1505

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

(Fig. S1), and compared the effects on protein and RNA expression,as well as on RNA processing and trafficking. We foundno significant differences between the two introns in proteinexpression (Fig. 7A–C), in nascent and total reporter RNAexpression (Fig. 7D–F), in nucleocytoplasmic reporter mRNAexport (Fig. 7G), or in loading of reporter RNAs onto polysomes fortranslation (Fig. 7H). Thus, these two introns, originating from twoseparate genes of different species and having no obvious sequencesimilarity with one another, were fully interchangeable whencoupled with the nefm promoter and 3′-UTR.

DISCUSSIONIn testing the activity of cis-regulatory elements of the nefm gene inlive Xenopus embryos, we found that incorporating the last intron ofnefm (Intron II) was essential for achieving optimal reporter proteinexpression. From analyzing the effects of various intron-bearingand intronless constructs on protein expression, and on thetranscription and handling of the RNAs, we conclude that theseeffects are due to post-transcriptional events requiring both splicingand internal elements of the intron rather than to effects ontranscription arising from intronic enhancer elements. The analysesof the handling of the RNAs further demonstrated that differences inprotein expression were contingent upon post-transcriptionalregulation operating cumulatively at multiple control points, butmost chiefly on nucleocytoplasmic export. These conclusions areespecially relevant for neurofilament genes because post-transcriptional events underlie many of the changes inneurofilament protein and RNA expression that promote axondevelopment (Moskowitz and Oblinger, 1995; Schwartz et al.,1994) and regeneration (Ananthakrishnan et al., 2008), and alsocontribute to neurodegenerative diseases (Ge et al., 2003).Introns have been known since the late 1970s to exert regulatory

effects on gene expression (Le Hir et al., 2003). Differences inexpression between intron-containing genes and their intronlesscounterparts have been demonstrated in multiple species, rangingacross plants (Bourdon et al., 2001; Callis et al., 1987), invertebrates(Duncker et al., 1997) and vertebrates (Braddock et al., 1994;Buchman and Berg, 1988; Heim et al., 2014; Matsumoto et al.,1998; Palmiter et al., 1991), including mammalian cell lines (Nottet al., 2003). Moreover, these influences can operate on bothtranscriptional (Brinster et al., 1988; McKenzie and Brennan, 1996;Sleckman et al., 1996) and post-transcriptional control. Post-transcriptional effects extend beyond those associated with nascentRNA splicing to include a full cascade of events associated withRNA metabolism, including polyadenylation, nucleocytoplasmicexport, translational efficiency and cytoplasmic decay (Le Hir et al.,2001b; Lewis and Izaurralde, 1997; Liu and Mertz, 1995; Luo andReed, 1999; Reed and Hurt, 2002; Singh et al., 2012; Vagner et al.,2000). Because these splicing-mediated effects on RNA regulationcan differ dramatically among individual intron sequences, theirsurrounding exon contents, the relative positioning of introns withina gene and the experimental systems in which they are tested, nogeneric mechanism has been found to account for all varieties ofeffects (Le Hir et al., 2003; Nott et al., 2003). Hence, each gene mustbe considered separately, and in the context of the cells and tissuesin which it is normally expressed.Most of the attention paid to the role of introns in the nervous

system has been in regard to alternative splicing (de la Grange et al.,2010; Grosso et al., 2008; Zheng and Black, 2013). This layer ofregulation contributes considerably to the diversity of neuronalmRNAs and their translated proteins, adding to the dynamic controlof neurogenesis, cell migration, and the formation, maturation and

function of synapses (Norris and Calarco, 2012). However, becauseno type IV neurofilament gene is known to undergo alternativesplicing, the role intron splicing plays for these genes has up to nowgone unstudied. Our study points to a role for introns ofneurofilament genes extending beyond alternative splicing toencompass important downstream gene regulatory functions thatare crucial for expression.

Results from our experiments with different endogenous andmodified introns indicate that, beyond simply having a spliceableintron, what was spliced mattered for subsequent stages of post-transcriptional regulation. Each intronic construct resulted instatistically significant effects on steady-state total RNA levels,but not on nascent RNA transcription, indicative of modest butnonetheless measurable effects on RNA processing. However, farand away the most marked effect was on nucleocytoplasmic exportof the mRNAs, which could readily have accounted for the vastdifferences in protein expression seen among the various nefmintron-bearing constructs. These experiments also demonstrated thatnot only did the two endogenous nefm full-length introns varymarkedly in their ability to promote (or alternatively, inhibit)nuclear export but also that sequences internal to Intron II wereneeded because only the full-length but not the minimal spliceableversion of this intron led to efficient export. Although the genericimportance of intron splicing and the varying abilities of particularintrons for promoting nucleocytoplasmic mRNA export have beenknown for some time (Le Hir et al., 2001b; Luo and Reed, 1999;Reed and Hurt, 2002), our results delineate specific parameters forintron splicing that are most relevant for the nefm gene. Theseparameters differ from those of other genes, such as oskar, forwhich splicing at the position of the first of three exons is requiredfor nuclear export and subsequent localization of the mRNA to theposterior pole of the Drosophila embryo, and the third intron cansubstitute for the first, as long as it is placed into the position of thefirst intron within the gene (Hachet and Ephrussi, 2004). At thispoint, we can only speculate whether the observed differencesbetween the two endogenous nefm introns might confer anyadvantages on neurons. One possibility is that splicing of the firstintron signals that transcription and processing are as yetincomplete, acting as part of a surveillance mechanism to preventexport and translation of prematurely truncated transcripts (Yapet al., 2015). Translation of such transcripts would be particularlydeleterious for NEFM because such truncated transcripts cangenerate dominant-negative proteins to disrupt pre-existingfilaments with severe consequences for the neuron (Lin andSzaro, 1996; Wong and Cleveland, 1990).

Parsing the molecular mechanisms responsible for conferringthese optimal properties on the full-length Intron II goes beyond thescope of the current study, for they are undoubtedly complex (Reed,2003). Possibilities include: (1) influences of exonic sequencesadjoining the intron, which were preserved in our constructs andwhich can sometimes affect deposition of the exon junctioncomplex (Le Hir et al., 2000, 2001b); (2) differences amongintrons in their associations with trans-factors, which cansubsequently be transferred to the spliced mRNA to regulatedownstream events (Hachet and Ephrussi, 2001, 2004; Le Hir et al.,2001a, 2003; Lehmann and Nusslein-Volhard, 1986; Micklemet al., 1997; Mohr et al., 2001); and (3) effects of intron length andtertiary structure on splicing efficiency, which can also affectdeposition of crucial trans-factors (Ohno et al., 2002). It seems, atleast, that length alone cannot be the whole answer because,although the optimally functional nefm Intron II (2.9 kb) wasnotably longer than the non-functional nefm Intron I (0.9 kb) and

1506

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

mini spliceable Intron II (0.2 kb), the equally functional hbb2 IntronB (0.6 kb) was actually slightly shorter than nefm Intron I. Thus,differences related to either the tertiary structures of the intron and

spliced transcript or the ability of RNA-binding proteins torecognize sequences within the introns seem just as likely tocontribute to the effects, if not more so, than intron length alone.

Fig. 7. Post-transcriptional regulation occurred in the same manner for nefm Intron II and hbb2 Intron B when each was paired with the nefm 3′-UTR,leading to equivalent amounts of protein expression for both. Embryos were co-injected with the same reference plasmid as described in Fig. 2 and a new testplasmid that contained the hbb2 Intron B and the nefm 3′-UTR. Reporter protein and RNA expression from this construct were then compared with those fromthe nefm Intron II construct. (A–C) The hbb2 Intron B promoted DsRed2 protein expression at levels comparable to that of nefm Intron II. Immunohistochemistry(A, hbb2 Int B; Fig. 2, nefm Intron II) and quantitative analysis by fluorescent SDS PAGE (B,C) were performed as described in Fig. 2. Dotted lines outline the centralnervous system. Vertical dotted lines separate the injected and uninjected sides. Scale bars: 20 µm (hindbrain); 50 µm (spinal cord). (D–H) Analyses of variousaspects of RNA regulation showed identical post-transcriptional processing of both RNAs. Analyses in D–Gwere performed as described in Fig. 4A–D, respectively.(Note, analyses of glGFP reference mRNA processing and nucleocytoplasmic export are shown in Fig. S3, and showed no significant differences across thesegroups). (H) RNAs arising from both intron groups were also actively loaded onto polysomes for translation to similar extents, as determined from polysomal profilingon sucrose gradients. Left axis: amount of total RNA (dotted line) in each fraction (A260). Fractions occupied by untranslated RNAs (pre), monosomes (mono) andpolysomes (poly) are as indicated. Right axis: percentage of glGFP (green bars) and DsRed2 (magenta bars) reporter RNAs present in the gradient. Error bars ares.e.m.; n=3 replicates of 30 pooled embryos (C-F) or n=3 biological replicates of ten embryos each (G), for each condition. N.S., P>0.05 (Student’s t-test).

1507

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

At the moment, such properties are difficult to predict solely byusing bioinformatics, because of its limited ability to determinefolded RNA structures, and because recognition sequences forRNA-binding proteins tend to be short and degenerate (Lagier-Tourenne et al., 2012; Thisted et al., 2001; Tollervey et al., 2011).Hence, the use of an experimental system, such as the one describedhere using Xenopus, will be essential for performing such studies.In demonstrating the usefulness of Xenopus embryos to analyze

the contributions of multiple gene regulatory events in the intactorganism, our study underscores the importance of intron splicing,extending its role beyond generating mRNA diversity to encompassmultiple downstream post-transcriptionally regulated eventsthat culminate in protein expression. Such studies are aimedultimately at gaining a better understanding of the post-transcriptional regulation of the endogenous nefm gene, becausesuch regulation, along with that of the other neurofilament genes, iscrucial for axon development and contributes to the pathogenicity ofneurodegenerative disorders. We are thus optimistic that thisexperimental system will provide future insights into the cis-regulatory features and associated trans-factors controlling thisregulation in vivo. The interchangeable nature of the last introns oftwo very distinct genes, originating from phylogenetically disparateorganisms, further indicates that these properties are both sharedamong multiple genes and conserved during evolution. Thus, thelessons learned are likely to define a class of mechanisms thatapplies more broadly to additional genes.

MATERIALS AND METHODSPreparation of plasmids for microinjectionTo obtain promoter-specific reporter expression in F0 stage 43–44 embryos,we used a modified expression plasmid, which incorporated an attB elementand two insulators flanking the reporter gene (Wang and Szaro, 2015). Formaking the DsRed2 test plasmids used to analyze effects of introns onexpression, we first replaced the rabbit β-globin (hbb2) 3′-UTR inpSPORT1[attB/Ins1/(1.5 kb) NF-M promoter/DsRed2/β-globin 3′-UTR/Ins2] (Wang and Szaro, 2015) with the Xenopus nefm (NF-M) 3′-UTR.Introns were then cloned into the XhoI site of this construct, which waslocated within the multiple cloning site of the original DsRed2 sequences.This site is downstream of the fluorescent protein itself but upstream of thestop codon. Introns were obtained by performing PCR using primers thatincluded nucleotide sequences both upstream of the 5′ and downstream ofthe 3′ splice sites (5′: 7, 11 and 5 nucleotides; 3′: 20, 17 and 10 nucleotides,for nefm Intron II, nefm Intron I and hbb2 Intron B, respectively) to preservethe contexts of the introns. Doing so resulted in a few amino acids (9, 9 and5, respectively) from the parent protein being retained at the C-terminus ofthe expressed DsRed2. The mini spliceable Intron II construct retained the98 and 100 nucleotides from the 5′ and 3′ ends of Intron II, respectively, andthe unspliceable Intron II construct lacked 11 and 21 nucleotides from the 5′and 3′ ends of the intron, respectively. For a complete list of oligonucleotideprimers [Integrated DNATechnologies (IDT), Coralville, IA] used to isolatethe various elements incorporated into the plasmids and for details oncloning each plasmid, see Table S1. The nefm reference plasmid waspSPORT1[attB/Ins1/(1.5 kb) NF-M promoter/glGFP/β-globin 3′-UTR/Ins2], as previously described (Wang and Szaro, 2015). Plasmid DNAswere amplified in Subcloning Efficiency DH5α Competent Cells (LifeTechnologies), extracted (Wizard Plus Miniprep, Promega; Plasmid Maxikit, Qiagen) and further purified (Wizard SV Gel and PCR Clean-UpSystem, Promega) prior to injection.

Microinjection of plasmid DNAs into embryosAll procedures involving animals were approved by the University atAlbany’s Institutional Animal Care and Use Committee. Fertilized eggswere obtained from breeding pairs of periodic albino (ap/ap) Xenopus laevis(Hoperskaya, 1975; Tompkins, 1977) through injection of human chorionicgonadotropin (CG-10; Sigma-Aldrich) into the dorsal lymph sac to induce

amplexus (Gurdon, 1967). Test and reference plasmid DNAs (75 pg each, in10 nl) were pressure-co-injected into a single blastomere of two-cell-stageembryos, as described previously (Gervasi and Szaro, 2004). Because inXenopus the descendants of each blastomere in spinal cord, hindbrain andadjacent somites are restricted to one side of the midline (Jacobson andHirose, 1978), comparing fluorescence between the two sides aided indistinguishing true expression from background autofluorescence,especially for poorly expressing plasmids.

Immunohistochemistry and microscopyStage 43–44 tadpoles were anesthetized and fixed, and transverse frozensections were cut through the hindbrain and spinal cord for immunostaining,as previously described (Wang and Szaro, 2015). The primary antibodiesused to detect glGFP and DsRed2 were, respectively: (1) anti-GFP antibodygenerated in goat (catalog number 600-101-215, lot number 18875,Rockland), 1:500; (2) anti-DsRed antibody generated in rabbit (catalognumber 632496, lot number 1306037, Living Colors DsRed polyclonalantibody, Clontech Laboratories), 1:500. Alexa-Fluor-488- (catalog numberA11055, lot number 757104, Life Technologies) and Alexa-Fluor-546-conjugated (catalog number A10040, lot number 1218269, LifeTechnologies) secondary antibodies against the appropriate species werethen used (1:1000) to detect the relevant fluorophores. Immunostainedsections were imaged by confocal laser scanning microscopy (Zeiss LSM510 or LSM 710; 20× Plan-ApoChromat; N.A., 0.75; Carl Zeiss AG).

Quantification of reporter protein expression by fluorescent SDSPAGETo quantify expression of the fluorescent reporter proteins, we adapted astrategy from bacterial studies for visualizing the fluorescent proteinsextracted from embryos, directly on partially denaturing SDS 12%polyacrylamide gels (Topilina et al., 2015). For each group, 30 stage 43–44 embryos were homogenized (Polytron 3000, Kinematica AG,Switzerland) in three volumes of lysis buffer [50 mM Tris-HCl, pH 8.0;10% (vol/vol) glycerol], and the homogenate was spun for 15 min (4°C) in amicrocentrifuge (18,400 g) to remove cellular debris. For optimalfluorescence quantification, two-embryo-equivalents of material were thenused for each gel. To maintain conditions mild enough to preserve reporterprotein fluorescence, the samples were mixed just prior to loading with aloading dye [10% (wt/vol) SDS, 10% (vol/vol) glycerol, 12.5% (vol/vol)stacking buffer, 0.1% bromophenol blue] but without boiling or addingβ-mercaptoethanol. Gels were run (Laemmli, 1970) and then imaged forfluorescence immediately afterwards to minimize fluorescence quenching(excitation at 488 nm and 526 nm for glGFP and DsRed2, respectively;Typhoon 9400 scanner, GE Healthcare). All related figures (Figs. 2, 3 and5–7) show an image of the full gel in color, to visualize coexpression ofglGFP and DsRed2 in the same lane (DsRed2 ran as a ∼54 kDa dimer andglGFP as a ∼27 kDa monomer), and separate monochrome images,representative of those used for quantification. The intensity of theDsRed2 band was normalized against that of glGFP in the same lane(NIH ImageJ64), and then averaged (mean±s.e.m.) over three replicate gelsfor each group. Statistical comparisons across three or more groups wereperformed using one-way ANOVA, with Tukey’s post-hoc tests foridentifying significant differences among individual groups. Comparisonsbetween two groups were performed using two-tailed Student’s t-tests.Differences were considered statistically significant at P<0.05. Whether ornot differences were significant, power analysis (Faul et al., 2007) was usedto confirm that sufficient numbers of samples had been analyzed to supportthe conclusions (Power>0.8).

Analyses of total RNA expression, nascent RNA expression,mRNA processing, nucleocytoplasmic transport and translationTo maximize recovery of long intron-bearing heterogeneous nuclear (hn)RNAs (Ananthakrishnan et al., 2008), total cellular RNAwas obtained from30 stage 43–44 embryos from each group by performing guanidine-isothiocyanate–CsCl ultracentrifugation (Davis et al., 1994). ResuspendedRNA pellets were treated with DNase I (RQ1 RNase-Free DNase, Promega)to further remove any residual DNA contaminant, and the RNAwas reverse

1508

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

transcribed (SuperScript III, Invitrogen) using a mixture of Oligo dT(Invitrogen) and gene-specific primer for DsRed2 (Table S1), each at2 pmol.

For analyzing levels of reporter RNA expression (which included bothprocessed mRNAs and nascent unspliced transcript), 5% of the resultantcDNA was subjected to RT-qPCR (7900HT Fast Real-Time PCR System,Life Technologies), using primers from the reporter RNA-coding sequences(Table S1) and SYBR Green PCRMaster Mix (Life Technologies).DsRed2expression relative to that of glGFP (ΔCT) in the same sample was averagedfrom three replicates for each group (mean±s.e.m.), and statistical analyseswere performed using the same tests and criteria for significance as foranalyses of protein expression. For presentation purposes (Figs 4A and 7D),results were first converted to relative expression levels (2–ΔCT) and thengraphed as a percentage of the values for the intronless constructs. Toconfirm there were no mutations in the resulting spliced RNA generated bythe injected embryos, we examined the sequences of the RNA derived fromevery group by amplifying the above cDNA with primers located in theflanking regions (forward in DsRed2 coding sequence: 5′-GTGATGAA-CTTCGAGGACG-3′; reverse in nefm 3′-UTR: 5′-GTAGTATTACAGT-GTCTGGAG-3′), cloning the resultant amplicon into pGEM-T Easy VectorSystems (Promega) by TA-cloning, and sequencing the plasmids extractedfrom three colonies for each group (GENEWIZ).

For analyzing nascent RNA expression, RT-qPCR was performed in asimilar manner on 5% of the same cDNA as described above, but with aforward primer located near to the 3′ end of each intron and a reverse primerfrom the 5′ end of the downstream 3′-UTR to detect only the unsplicednascent transcript. All primers were designed to generate amplicons of equallength across the groups compared (Table S1). Results for DsRed2 werenormalized against those for glGFP present in each sample (ΔCT), averagedover the samples from the group (mean±s.e.m.), and then converted to afold-difference (2–ΔCT; i.e. DsRed2/glGFP) for presentation. Statisticalanalyses were performed as described for total RNA expression.

mRNA processing efficiency was calculated by taking the ratio of theamount of nascent reporter RNA to that of total reporter RNA for each sampleand subtracting this ratio from 100% [ΔCT=CT(nascent RNA)–CT(total RNA);mRNA processing=100%–2–ΔCT]. Samples were averaged and statisticalanalyses were performed as above. As a control for the quality of the samples,glGFP RNA (Figs S2A and S3A) was analyzed separately from DsRed2RNA (Figs 4C and 7F) and was appropriately found to be processed in anequivalent manner across all groups.

For measuring nucleocytoplasmic export, subcellular fractionation wasperformed by homogenizing ten stage 43–44 tadpoles in 500 µl ofPolysomal Buffer A (25 mM NaCl, 5 mM MgCl2, 25 mM Tris-HClpH 7.4; 4°C) containing 10 mM of vanadyl ribonucleoside complex (NewEngland Biolabs) with glass homogenizers and centrifuging the sample atlow speed (285 g; 4°C) for 3 min to pellet nuclei (Ananthakrishnan et al.,2008). mRNA was extracted from each fraction (RNeasy Plus kit, Qiagen)and treated as above with DNase I to remove any potentially contaminatingDNA. RT-qPCR was then performed to assay reporter RNA expression, asabove, with primers targeting reporter coding sequences (Table S1). Foreach sample, ΔCT was then determined between cytoplasmic and nuclearRNAs, averaged across biological replicates (n=3 of 10 embryos each), andanalyzed for statistical significance as above. The ΔCT (mean±s.e.m.) wasthen converted to a fold change (2–ΔCT) to yield the ratio of the amount ofRNA in the cytoplasm/nucleus. As was done for the RNA processing data,glGFP RNA (Figs S2B and S3B) was analyzed separately from DsRed2RNA (Figs 4D and 7G) and found to be processed equivalently across allgroups.

For polysomal profiling, 500 µl of cytoplasmic fraction, prepared asabove, was layered onto a 10-ml linear 5–56% (wt/wt) sucrose gradient(Gradient Master 107, BioComp Instruments, Canada) in Polysomal BufferA and spun in an ultracentrifuge (169,000 g, 4°C, 2 h; SW41 rotor,Beckman Coulter, Pasadena, CA). Fractions (500 µl) were collected (AutoDensi-Flow Density Gradient Fractionator, Labconco, Kansas City, MO,USA), and the macromolecules were precipitated [50 µl 3M sodium acetate,20 µl 0.5 M Na2(EDTA), 2 µl linear acrylamide (Ambion, LifeTechnologies), 1.5 ml ethanol], resuspended in 200 µl of buffer (0.5%SDS, 0.1 M NaCl, 1 mM EDTA, 10 mM Tris, pH 7.4), and nucleic acids

were isolated by performing phenol–chloroform extraction and ethanolprecipitation (Davis et al., 1994). The pellets were then resuspended in 50 µlof reaction buffer for DNase I treatment, as above. After an additional roundof phenol–chloroform extraction and ethanol precipitation, the final RNAwas resuspended in 20 µl water, and 2 µl was used for quantification at A260

(NanoDrop ND1000, Thermo Scientific) to distinguish pre-monosomal,monosomal and polysomal fractions. Another 9 µl was used for RT-qPCRwith appropriate primers that yielded amplicons of equal length for allRNAs compared (Table S1), and statistical analysis was performed asdescribed for the RNA expression analyses.

AcknowledgementsWe thank Enrique Amaya (University of Manchester, UK), Richard Harland(University of California, Berkeley, CA) and Daniel Weeks (University of Iowa, IA) forthe tubb2b and actc1 promoters, the hbb2 3′-UTR and Intron B, and the originalattB–HS4 plasmids, respectively. We also thank Rupa Priscilla and ChristineGervasi for their help with Xenopus spawnings, Min-Ho Lee for providing access toequipment for polysomal profiling and Erica Hutchins for helpful editorial comments.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsC.W. conceived, designed and performed the experiments, analyzed andinterpreted the data, and wrote the manuscript. B.S. conceived of experiments,interpreted data and wrote the manuscript.

FundingFinancial support for this work came from the National Science Foundation –

Division of Integrative Organismal Systems [grant number IOS 1257449 to B.G.S.];and a pre-doctoral fellowship from the American Association of University Women(International Fellow, C.W.).

Supplementary informationSupplementary information available online athttp://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.185199/-/DC1

ReferencesAnanthakrishnan, L. and Szaro, B. G. (2009). Transcriptional and translational

dynamics of light neurofilament subunit RNAs during Xenopus laevis optic nerveregeneration. Brain Res. 1250, 27-40.

Ananthakrishnan, L., Gervasi, C. and Szaro, B. G. (2008). Dynamic regulation ofmiddle neurofilament RNA pools during optic nerve regeneration. Neuroscience153, 144-153.

Arai, T., Hasegawa, M., Akiyama, H., Ikeda, K., Nonaka, T., Mori, H., Mann, D.,Tsuchiya, K., Yoshida, M., Hashizume, Y. et al. (2006). TDP-43 is a componentof ubiquitin-positive tau-negative inclusions in frontotemporal lobar degenerationand amyotrophic lateral sclerosis. Biochem. Biophys. Res. Commun. 351,602-611.

Bennett, G. S., Hollander, B. A. and Laskowska, D. (1988). Expression andphosphorylation of the mid-sized neurofilament protein NF-M during chick spinalcord neurogenesis. J. Neurosci. Res. 21, 376-390.

Bourdon, V., Harvey, A. and Lonsdale, D. M. (2001). Introns and their positionsaffect the translational activity of mRNA in plant cells. EMBO Rep. 2, 394-398.

Braddock, M., Muckenthaler, M., White, M. R. H., Thorbum, A. M., Sommerville,J., Kingsman, A. J. and Kingsman, S. M. (1994). Intron-less RNA injected intothe nucleus ofXenopus oocytes accesses a regulated translation control pathway.Nucleic Acids Res. 22, 5255-5264.

Brinster, R. L., Allen, J. M., Behringer, R. R., Gelinas, R. E. and Palmiter, R. D.(1988). Introns increase transcriptional efficiency in transgenic mice. Proc. Natl.Acad. Sci. USA 85, 836-840.

Buchman, A. R. and Berg, P. (1988). Comparison of intron-dependent and intron-independent gene expression. Mol. Cell. Biol. 8, 4395-4405.

Callis, J., Fromm, M. and Walbot, V. (1987). Introns increase gene expression incultured maize cells. Genes Dev. 1, 1183-1200.

Canete-Soler, R., Silberg, D. G., Gershon, M. D. and Schlaepfer, W. W. (1999).Mutation in neurofilament transgene implicates RNA processing in thepathogenesis of neurodegenerative disease. J. Neurosci. 19, 1273-1283.

Cao, W., Razanau, A., Feng, D., Lobo, V. G. and Xie, J. (2012). Control ofalternative splicing by forskolin through hnRNP K during neuronal differentiation.Nucleic Acids Res. 40, 8059-8071.

Carden, M. J., Trojanowski, J. Q., Schlaepfer, W. W. and Lee, V. M. Y. (1987).Two-stage expression of neurofilament polypeptides during rat neurogenesiswith early establishment of adult phosphorylation patterns. J. Neurosci. 7,3489-3504.

1509

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

Davis, L. G., Kuehl, W. M. and Battey, J. F. (1994). Basic Methods in MolecularBiology. Norwalk: Appleton and Lange.

de la Grange, P., Gratadou, L., Delord, M., Dutertre, M. and Auboeuf, D. (2010).Splicing factor and exon profiling across human tissues. Nucleic Acids Res. 38,2825-2838.

Duncker, B. P., Davies, P. L. and Walker, V. K. (1997). Introns boost transgeneexpression in Drosophila melanogaster. Mol. Gen. Genet. 254, 291-296.

Faul, F., Erdfelder, E., Lang, A.-G. and Buchner, A. (2007). G*Power 3: a flexiblestatistical power analysis program for the social, behavioral, and biomedicalsciences. Behav. Res. Methods 39, 175-191.

Ge, W.-W., Leystra-Lantz, C., Wen, W. and Strong, M. J. (2003). Selective loss oftrans-acting instability determinants of neurofilament mRNA in amyotrophic lateralsclerosis spinal cord. J. Biol. Chem. 278, 26558-26563.

Gervasi, C. and Szaro, B. G. (2004). Performing functional studies of Xenopuslaevis intermediate filament proteins through injection of macromolecules intoearly embryos. Methods Cell Biol. 78, 673-701.

Grosso, A. R., Gomes, A. Q., Barbosa-Morais, N. L., Caldeira, S., Thorne,N. P., Grech, G., von Lindern, M. and Carmo-Fonseca, M. (2008). Tissue-specific splicing factor gene expression signatures. Nucleic Acids Res. 36,4823-4832.

Gurdon, J. B. (1967). African clawed frogs. In Methods in Developmental Biology(ed. F. H. Wilt and N. K. Wessels), pp. 75-84. New York: T.Y.Crowell.

Hachet, O. and Ephrussi, A. (2001). Drosophila Y14 shuttles to the posterior ofthe oocyte and is required for oskar mRNA transport. Curr. Biol. 11,1666-1674.

Hachet, O. and Ephrussi, A. (2004). Splicing of oskar RNA in the nucleus iscoupled to its cytoplasmic localization. Nature 428, 959-963.

Hamer, D. H., Smith, K. D., Boyer, S. H. and Leder, P. (1979). SV40 recombinantscarrying rabbit beta-globin gene coding sequences. Cell 17, 725-735.

Heim, A. E., Hartung, O., Rothhamel, S., Ferreira, E., Jenny, A. andMarlow, F. L.(2014). Oocyte polarity requires a Bucky ball-dependent feedback amplificationloop. Development 141, 842-854.

Hoperskaya, O. A. (1975). The development of animals homozygous for a mutationcausing periodic albinism (ap) in Xenopus laevis. J. Embryol. Exp. Morphol. 34,253-264.

Jacobson, M. and Hirose, G. (1978). Origin of the retina from both sides of theembryonic brain: a contribution to the problem of crossing at the optic chiasma.Science 202, 637-639.

Kwiatkowski, T. J., Jr., Bosco, D. A., LeClerc, A. L., Tamrazian, E., Vanderburg,C. R., Russ, C., Davis, A., Gilchrist, J., Kasarskis, E. J., Munsat, T. et al.(2009). Mutations in the FUS/TLS gene on chromosome 16 cause familialamyotrophic lateral sclerosis. Science 323, 1205-1208.

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of thehead of bacteriophage T4. Nature 227, 680-685.

Lagier-Tourenne, C. and Cleveland, D. W. (2009). Rethinking ALS: the FUS aboutTDP-43. Cell 136, 1001-1004.

Lagier-Tourenne, C., Polymenidou, M., Hutt, K. R., Vu, A. Q., Baughn, M.,Huelga, S. C., Clutario, K. M., Ling, S.-C., Liang, T. Y., Mazur, C. et al. (2012).Divergent roles of ALS-linked proteins FUS/TLS and TDP-43 intersect inprocessing long pre-mRNAs. Nat. Neurosci. 15, 1488-1497.

Le Hir, H., Izaurralde, E., Maquat, L. E. and Moore, M. J. (2000). The spliceosomedeposits multiple proteins 20–24 nucleotides upstream of mRNA exon-exonjunctions. EMBO J. 19, 6860-6869.

Le Hir, H., Gatfield, D., Braun, I. C., Forler, D. and Izaurralde, E. (2001a). Theprotein Mago provides a link between splicing and mRNA localization. EMBORep. 2, 1119-1124.

Le Hir, H., Gatfield, D., Izauuralde, E. and Moore, M. J. (2001b). The exon-exonjunction complex provides a binding platform for factors involved in mRNA exportand nonsense-mediated mRNA decay. EMBO J. 20, 4987-4997.

Le Hir, H., Nott, A. and Moore, M. J. (2003). How introns influence and enhanceeukaryotic gene expression. Trends Biochem. Sci. 28, 215-220.

Lehmann, R. and Nusslein-Volhard, C. (1986). Abdominal segmentation, pole cellformation, and embryonic polarity require the localized activity of oskar, a maternalgene in Drosophila. Cell 47, 141-152.

Lewis, J. D. and Izaurralde, E. (1997). The role of the cap structure in RNAprocessing and nuclear export. Eur. J. Biochem. 247, 461-469.

Lin, W. and Szaro, B. G. (1996). Effects of intermediate filament disruption on theearly development of the peripheral nervous system of Xenopus laevis. Dev. Biol.179, 197-211.

Liu, X. andMertz, J. E. (1995). HnRNP L binds a cis-acting RNA sequence elementthat enables intron-dependent gene expression. Genes Dev. 9, 1766-1780.

Liu, Y. and Szaro, B. G. (2011). hnRNP K post-transcriptionally co-regulatesmultiple cytoskeletal genes needed for axonogenesis. Development 138,3079-3090.

Liu, Y., Gervasi, C. and Szaro, B. G. (2008). A crucial role for hnRNP K in axondevelopment in Xenopus laevis. Development 135, 3125-3135.

Luo, M.-J. and Reed, R. (1999). Splicing is required for rapid and efficient mRNAexport in metazoans. Proc. Natl. Acad. Sci. USA 96, 14937-14942.

Matsumoto, K., Wassarman, K. M. and Wolffe, A. P. (1998). Nuclear history of apre-mRNA determines the translational activity of cytoplasmic mRNA. EMBO J.17, 2107-2121.

McKenzie, R. W. and Brennan, M. D. (1996). The two small introns of theDrosophila affinidisjuncta Adh gene are required for normal transcription. NucleicAcids Res. 24, 3635-3642.

Micklem, D. R., Dasgupta, R., Elliott, H., Gergely, F., Davidson, C., Brand, A.,Gonzalez-Reves, A. and St Johnston, D. (1997). The mago nashi gene isrequired for the polarisation of the oocyte and the formation of perpendicular axesin Drosophila. Curr. Biol. 7, 468-478.

Mohr, S. E., Dillon, S. T. and Boswell, R. E. (2001). The RNA-binding proteinTsunagi interacts with Mago Nashi to establish polarity and localize oskar mRNAduring Drosophila oogenesis. Genes Dev. 15, 2886-2899.

Moskowitz, P. F. and Oblinger, M. M. (1995). Transcriptional and post-transcriptional mechanisms regulating neurofilament and tubulin geneexpression during normal development of the rat brain. Mol. Brain Res. 30,211-222.

Moujalled, D., James, J. L., Yang, S., Zhang, K., Duncan, C., Moujalled, D. M.,Parker, S. J., Caragounis, A., Lidgerwood, G., Turner, B. J. et al. (2015).Phosphorylation of hnRNP K by cyclin-dependent kinase 2 controls cytosolicaccumulation of TDP-43. Hum. Mol. Genet. 24, 1655-1669.

Nie, Z., Wu, J., Zhai, J., Lin, H., Ge, W., Schlaepfer, W. W. and Canete-Soler, R.(2002). Untranslated element in neurofilament mRNA has neuropathic effect onmotor neurons of transgenic mice. J. Neurosci. 22, 7662-7770.

Norris, A. D. and Calarco, J. A. (2012). Emerging roles of alternative pre-mRNAsplicing regulation in neuronal development and function. Front. Neurosci. 6, 122.

Nott, A., Meislin, S. H. and Moore, M. J. (2003). A quantitative analysis of introneffects on mammalian gene expression. RNA 9, 607-617.

Ohno, M., Segref, A., Kuersten, S. andMattaj, I. W. (2002). Identity elements usedin export of mRNAs. Mol. Cell 9, 659-671.

Palmiter, R. D., Sandgren, E. P., Avarbock, M. R., Allen, D. D. and Brinster, R. L.(1991). Heterologous introns can enhance expression of transgenes in mice.Proc. Natl. Acad. Sci. USA 88, 478-482.

Rafiq, M., Suen, C. K. M., Choudhury, N., Joannou, C. L., White, K. N. andEvans, R. W. (1997). Expression of recombinant human ceruloplasmin – anabsolute requirement for splicing signals in the expression cassette. FEBS Lett.407, 132-136.

Reed, R. (2003). Coupling transcription, splicing and mRNA export. Curr. Opin. CellBiol. 15, 326-331.

Reed, R. and Hurt, E. (2002). A conserved mRNA export machinery coupled to pre-mRNA splicing. Cell 108, 523-531.

Roosa, J. R., Gervasi, C. and Szaro, B. G. (2000). Structure, biological activity ofthe upstream regulatory sequence, and conserved domains of a middle molecularmass neurofilament gene of Xenopus laevis. Mol. Brain Res. 82, 35-51.

Schwartz, M. L., Shneidman, P. S., Bruce, J. and Schlaepfer, W. W. (1990).Axonal dependency of the postnatal upregulation in neurofilament expression.J. Neurosci. Res. 27, 193-201.

Schwartz, M. L., Shneidman, P. S., Bruce, J. and Schlaepfer, W. W. (1994).Stabilization of neurofilament transcripts during postnatal development.Mol. BrainRes. 27, 215-220.

Singh, G., Kucukural, A., Cenik, C., Leszyk, J. D., Shaffer, S. A., Weng, Z. andMoore, M. J. (2012). The cellular EJC interactome reveals higher-order mRNPstructure and an EJC-SR protein nexus. Cell 151, 750-764.

Sleckman, B. P., Gorman, J. R. and Alt, F. W. (1996). Accessibility control ofantigen-receptor variable-region gene assembly: role of cis-acting elements.Annu. Rev. Immunol. 14, 459-481.

Strong, M. J., Volkening, K., Hammond, R., Yang, W., Strong, W., Leystra-Lantz, C. and Shoesmith, C. (2007). TDP43 is a human low molecular weightneurofilament (hNFL) mRNA-binding protein. Mol. Cell. Neurosci. 35,320-327.

Sun, X. and Maquat, L. E. (2000). mRNA surveillance in mammalian cells: therelationship between introns and translation termination. RNA 6, 1-8.

Szaro, B. G. and Gainer, H. (1988). Identities, antigenic determinants, andtopographic distributions of neurofilament proteins in the nervous systems of adultfrogs and tadpoles of Xenopus laevis. J. Comp. Neurol. 273, 344-358.

Szaro, B. G. and Strong, M. J. (2010). Post-transcriptional control ofneurofilaments: new roles in development, regeneration and neurodegenerativedisease. Trends Neurosci. 33, 27-37.

Szaro, B. G. and Strong, M. J. (2011). Regulation of cytoskeletal composition inneurons: transcriptional and post-transcriptional control in development,regeneration, and disease. Adv. Neurobiol. 3, 559-602.

Szaro, B. G., Lee, V. M.-Y. and Gainer, H. (1989). Spatial and temporal expressionof phosphorylated and non-phosphorylated forms of neurofilament proteins in thedeveloping nervous system of Xenopus laevis. Brain Res. Dev. Brain Res. 48,87-103.

Thisted, T., Lyakhov, D. L. and Liebhaber, S. A. (2001). Optimized RNA targets oftwo closely related triple KH domain proteins, heterogeneous nuclearribonucleoprotein K and αCP-2KL, suggest distinct modes of RNA recognition.J. Biol. Chem. 276, 17484-17496.

1510

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience

Thyagarajan, A. and Szaro, B. G. (2004). Phylogenetically conserved binding ofspecific K Homology domain proteins to the 3′-untranslated region of thevertebrate middle neurofilament mRNA. J. Biol. Chem. 279, 49680-49688.

Thyagarajan, A. and Szaro, B. G. (2008). Dynamic endogenous association ofneurofilament mRNAs with K-homology domain ribonucleoproteins in developingcerebral cortex. Brain Res. 1189, 33-42.

Thyagarajan, A., Strong, M. J. andSzaro, B. G. (2007). Post-transcriptional controlof neurofilaments in development and disease. Exp. Cell Res. 313, 2088-2097.

Tollervey, J. R., Curk, T., Rogelj, B., Briese, M., Cereda, M., Kayikci, M., Konig,J., Hortobagyi, T., Nishimura, A. L., Župunski, V. et al. (2011). Characterizingthe RNA targets and position-dependent splicing regulation by TDP-43. Nat.Neurosci. 14, 452-458.

Tompkins, R. (1977). Grafting analysis of the periodic albino mutant of Xenopuslaevis. Dev. Biol. 57, 460-464.

Topilina, N. I., Green, C. M., Jayachandran, P., Kelley, D. S., Stanger, M. J.,Piazza, C. L., Nayak, S. and Belfort, M. (2015). SubB intein of Mycobacteriumtuberculosis as a sensor for oxidative and nitrosative stresses. Proc. Natl. Acad.Sci. USA 112, 10348-10353.

Undamatla, J. and Szaro, B. G. (2001). Differential expression and localization ofneuronal intermediate filament proteins within newly developing neurites in

dissociated cultures of Xenopus laevis embryonic spinal cord. Cell Motil.Cytoskeleton 49, 16-32.

Vagner, S., Vagner, C. andMattaj, I. W. (2000). The carboxyl terminus of vertebratepoly(A) polymerase interacts with U2AF 65 to couple 3′-end processing andsplicing. Genes Dev. 14, 403-413.

Wang, C. and Szaro, B. G. (2015). A method for using direct injection of plasmidDNA to study cis-regulatory element activity in F0Xenopus embryos and tadpoles.Dev. Biol. 398, 11-23.

Wong, P. C. and Cleveland, D. W. (1990). Characterization of dominant andrecessive assembly defective mutations in mouse neurofilament NF-M. J. CellBiol. 111, 1987-2003.

Yap, K., Lim, Z. Q., Khandelia, P., Friedman, B. and Makeyev, E. V. (2015).Coordinated regulation of neuronal mRNA steady-state levels throughdevelopmentally controlled intron retention. Genes Dev. 26, 1209-1223.

Zhao, Y. and Szaro, B. G. (1995). The optic tract and tectal ablation influence thecomposition of neurofilaments in regenerating optic axons of Xenopus laevis.J. Neurosci. 15, 4629-4640.

Zheng, S. and Black, D. L. (2013). Alternative pre-mRNA splicing in neurons:growing up and extending its reach. Trends Genet. 29, 442-448.

1511

RESEARCH ARTICLE Journal of Cell Science (2016) 129, 1500-1511 doi:10.1242/jcs.185199

Journal

ofCe

llScience